Failed GFCI?

While working around a receptacle outlet yesterday, I was rewarded with a loud POP when a hot terminal brushed the (steel) box. The screw head was slightly vaporized, with a matching pockmark in the box. The main 20A breaker tripped. Problem is, this was a downstream receptacle from an upstream GFCI, and the GFCI did not trip. I reset the main breaker, pressed the GFCI's TEST button, it tripped, and both it and its downstream receptacles went dead. It's a 2-month-old P&S GFCI made in (gulp!) China. The wiring all checks OK, and a 3-light tester shows all receptacles wired properly.

If I had been naked, standing in a bathtub full of water, and grabbed that receptacle hot side, I'd probably be dead. Is there any other explanation of what went wrong other than a failed GFCI? If it's bad, why does it test OK? These things are supposed to trip at 6ma within 25ms. I'm going to make up a GFCI tester I can plug in to test downstream receptacles individually and will post the results.

You can purchase a GFCI tester that will tell you if the GFCI is within specs. when it trips.

I would recomend buying one of them instead of making one your self.

There were a lot of GFCI recalled earlier this year. Probably chinese. Problem was they didn't trip. I will try to find the recall info. What brand are yours?

This recall is from 2003, but I think the same brand was in trouble eariler this year.>>
http://dbs.idaho.gov/electrical/Notices/gfci.pdf

Jimbo it is a P&S

Cass said:

I would recomend buying one of them instead of making one your self.

Click to expand...

Yeah, but then I'd have to get out of my jammies and go shopping . Guess I'll have to bite that bullet. I've e-mailed P&S tech support for advice also.

The GFCI is probably alright. What you had was a short to ground and the breaker did it's job quicker than the GFCI. If you had been standing in water and touched the hot wire and ground the difference in potential would have made the GFCI trip.

Do you know that the downstream circuit was connected to the protected load terminals of the GFCI breaker?

You get two choices when connecting a GFCI and one choice doesn't protect the downstream circuits.

Mikey said:

Yeah, but then I'd have to get out of my jammies and go shopping .

Click to expand...

Why get dressed?

GFCI tester

You learn something new every day

sbrn33 -- you got it mostly right. Here's P&S's response:

If the steel box was properly grounded, then yes, this definitely would have caused a differential current in the GFCI (very high current in the line leg, no return current in the neutral). If this was the case, then very simply, we think the circuit breaker just tripped out sooner than the GFCI did. A typical trip time for a magnetic circuit breaker with this type of overcurrent is 10ms or less. Our GFCI response is about 25ms.

Sounds like everything worked exactly as it should.

The GFCI's you have, based upon the tests you performed, are working properly.

Mike Gaines, Technical Support
Pass & Seymour/legrand

Bob -- yes, everything is connected properly.

Sparky -- I think they put you in jail for that in this county.

Mikey said:

Sparky -- I think they put you in jail for that in this county.

Click to expand...

For what? Buying from fleabay?

No, going to Lowe's in your jammies.

Mikey said:

No, going to Lowe's in your jammies.

Click to expand...

The link is for a GFI tester listed on Fleabay.

Oh, THAT's what it was

The link got trashed by the forum censor... I'll look there, though. There appear to be cheap testers and expensive testers. I assume the cheap ones just generate a 6ma load and see if things trip, whereas the expensive ones will tell you how long it took to trip as well? I made a half-*ssed attempt to find an internal diagram of a GFCI, but all of the images I found weren't available for one reason or another.

Found the Triplett Plug Bug on that site. They're cheaper at my local Lowe's (GB, not Triplett, but one of the "cheap kind" of testers).

There's a Greenlee 5708 there, which is expensive ($90) but I don't see why it's any better than the $7 ones. It's apparently the base device for a bunch of test sensors, but even the Greenlee Web site is pretty sparse in its description of the thing.